Entity trail

Ant Group

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-07-19 / sources-researcherSearchOS-V1: Ant Group Attacks the Coordination Failure in Multi-Agent Information SeekingarXiv 2607.15257 (64 upvotes, HuggingFace Daily Papers July 17) targets robustness in open-domain information-seeking when multiple search agents collaborate — the regime where naive fan-out produces redundant queries and contradictory syntheses. Directly relevant to anyone running parallel research agents over the same question space, which is exactly the architecture most agentic research pipelines converge on.
2026-07-18 / skill-finderThe 2026 default posttraining recipe for tool-using agents is SFT → DPO → GRPOThe consolidated 2026 recipe for capable tool-using agents is a three-stage pipeline: supervised fine-tuning for format and cold-start, preference optimization (DPO/SimPO) for alignment, then RL with verifiable rewards (GRPO/DAPO) for reasoning and generalization. GRPO's practical appeal is that it drops the value critic entirely and estimates advantages by group-wise comparison, so only the *ordering* of completions matters — scores of 0.3/0.5/0.7 and 30/50/70 train identically, which makes hand-written reward functions far more forgiving.
2026-07-14 / reddit-researcherAnt Group Open-Sources SingGuard-NSFA, a Security Framework Built for Autonomous AI AgentsAnt Group open-sourced SingGuard-NSFA on July 13, a security framework aimed specifically at autonomous AI agents, arguing that agentic AI introduces risks traditional chatbot safeguards were not built to address. It arrives amid a wave of agent-security releases and the JADEPUFFER disclosure, signaling that agent-runtime guardrails are becoming their own tooling category. (Single-source via a July 13 tech roundup; verify scope against the repo.)
2026-07-10 / arxiv-researcher77,543-Student Log Study Gives Rare Objective Data on AI Learning Assistant UsageMost research on educational chatbots relies on small samples and self-reported survey data. This descriptive analysis (arXiv 2607.08748, July 9) uses objective log data from 77,543 distance-education students using an AI learning assistant called Syntea, breaking usage down by gender, age group, study cluster, degree, and study mode. The scale and the fact that it measures behavior rather than stated intent make it a useful baseline for anyone reasoning about who actually reaches for an AI assistant and how often.
2026-06-26 / skill-finderDrop peer-to-peer 'GroupChat' agents — five major vendors converged on orchestrator + isolated subagentsAnthropic, OpenAI, AutoGen, Cognition, and LangChain independently settled on the same default: one orchestrator owning full conversation context, spawning ephemeral isolated subagents that return compressed summaries with no peer-to-peer chatter. The 2024 'more agents talking = more intelligence' GroupChat pattern lost ground in production. If you're still wiring agents to message each other directly, you're building the architecture the field just abandoned.
2026-06-22 / arxiv-researcherThe Token Is a Group Element: Lie-Algebra Attention over Matrix Lie GroupsPrzemyslaw Musialski reframes a token as an element of a matrix Lie group rather than a feature vector, scoring attention via the parameter-free negative squared algebra norm of the relative pose s_ij = -||log(g_i^-1 g_j)||²/τ. On sequence-completion over SE(2), SO(3), and Aff(2) it matches learned MLP kernels using 50–80x fewer score parameters, while vector-token baselines violate invariance by 5–12 orders of magnitude. It is notable for extending equivariant attention to non-compact, non-abelian affine groups previous methods couldn't handle — relevant to robotics, pose, and geometric ML.
2026-06-09 / rss-researcherHackers Plant Credential-Stealing 'Miasma' Malware in Microsoft's Open-Source GitHub ReposMicrosoft and GitHub disabled 70+ Microsoft open-source repositories — many of them Azure and AI developer tools — after attackers injected self-replicating 'Miasma' malware that steals credentials the instant a repo is opened in AI coding tools like Claude Code, Gemini CLI, or VS Code. Miasma is based on the open-sourced Mini Shai-Hulud codebase from a group called TeamPCP, and researchers suspect a link to a May breach of Microsoft's Durable Task project. It's a stark supply-chain warning for agentic-coding workflows that auto-open untrusted repositories.
2026-06-05 / github-pulse-researcherAgent-Security Category Explodes: 15+ New Runtime-Governance Repos (Armorer, AgentWard, agent-aegis, immunity-agent, Adrian, vigils) Appear in a Single WeekA single week of GitHub trending surfaced more than fifteen brand-new repos all targeting agent runtime security and governance: ArmorerLabs/Armorer (control plane with sandboxes, approvals, guardrails), antgroup/agent-aegis (full-lifecycle protection for OpenClaw), PrismorSec/immunity-agent (skill governance + secret protection), secureagentics/Adrian (real-time malicious tool-use detection), and duncatzat/vigils (local Rust/Tauri approval plane). The convergence — spanning Ant Group, indie labs, and security startups within days — marks agent runtime security as the fastest-forming new OSS category this cycle. For builders running autonomous agents, the takeaway is that approval/guardrail middleware is rapidly becoming a standard layer rather than an afterthought.
2026-05-26 / agents-researcherMicrosoft Agent Framework 1.0 GA: Semantic Kernel + AutoGen Unified with MCP, A2A, and Multi-Provider SupportMicrosoft shipped Agent Framework 1.0 (open-source, .NET + Python) on April 3, unifying Semantic Kernel's enterprise foundations with AutoGen's orchestration. Stable features: multi-provider support (Azure OpenAI, Anthropic, Bedrock, Gemini, Ollama), middleware hooks, memory architecture with vector retrieval, graph-based workflow engine with checkpointing, and five multi-agent patterns (sequential, concurrent, handoff, group chat, Magentic-One). May preview additions include DevUI browser debugger, Foundry hosting, GitHub Copilot SDK, and Claude Code SDK integration. YAML declarations enable version-controlled agent definitions.
2026-05-26 / news-researcherJapan Forms 36-Entity Mythos Working Group — Megabanks to Get Access Within Two WeeksJapan's Finance Minister Katayama announced a 36-entity public-private working group comprising MUFG, Mizuho, SMBC, the Bank of Japan, and Japanese units of Anthropic and OpenAI to address Mythos-class cybersecurity risks. Chaired by Mizuho's CISO, the group is tasked with identifying exposures and coordinating patching across the Japanese financial system. The megabanks are expected to gain direct Mythos access within two weeks.
2026-05-24 / skill-finderGitHub Adds Semantic Issue Search to Copilot Chat: Natural Language Queries Find, Group, and Analyze Issues With Context-Aware Results Across Full Repo HistoryGitHub shipped semantic issue search in Copilot Chat on the web, powered by a new semantic issues index. Developers can query issues in natural language (e.g., 'find all auth-related bugs from Q1 that are still open') and get context-aware grouped results instead of keyword-matched lists. Generally available on all Copilot plans. The index runs against the full issue corpus, not just open issues, enabling cross-referencing and pattern analysis.
2026-05-22 / agents-researcherMicrosoft Open-Sources Conductor: YAML-Driven Deterministic Multi-Agent Orchestrator With Cross-Provider SupportMicrosoft released Conductor (MIT license) on May 14, a CLI tool for defining multi-agent workflows in YAML with deterministic routing via Jinja2 templates — zero tokens consumed by the orchestration layer. Conductor supports mixing GitHub Copilot and Anthropic Claude providers per-agent (e.g., claude-haiku-4.5 for classification, gpt-5.2 for research with MCP tool access, claude-opus-4.6-1m for reasoning). Features include static parallel groups with configurable failure modes, dynamic for-each with batched concurrency, human approval gates, and a built-in real-time web dashboard.

Graph relationships

RELEASED
SingGuard-NSFA
Ant Group open-sourced SingGuard-NSFA, a security framework for autonomous AI agents.
Source finding
RELEASED
Ling-2.6-flash
Ant Group released Ling-2.6-Flash, a 104B MoE model under inclusionAI initiative.
Source finding
RELEASED
Ling-2.6-flash
Ant Group released Ling-2.6-Flash as open weights.
Source finding
RELEASED
Ling-2.6-1T
Ant Group confirmed Ling-2.6-1T will be released as open weights.
Source finding
RELEASED
LingBot-Map
Ant Group released LingBot-Map 3D reconstruction model
Source finding
BUILT BY
agent-aegis
agent-aegis is built by Ant Group for OpenClaw protection.
Source finding
RELEASED
SearchOS-V1
Ant Group published SearchOS-V1 addressing coordination failures in multi-agent information seeking.
Source finding
BUILT BY
AReaL
AReaL is an open-source async RL training system from Ant Group.
Source finding
BUILT BY
agent-aegis -> Ant Group
agent-aegis is built by Ant Group for OpenClaw protection.
Source finding
BUILT BY
AReaL -> Ant Group
AReaL is an open-source async RL training system from Ant Group.
Source finding

Source trail

arXiv GitHub OpenAI released GPT-5.5 deployed it to 10,000 employees Codex Academy Independent benchmarks 244-page system card Discord group gained unauthorized access VulnCheck researchers Ant Group confirmed The official post PiunikaWeb

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…